Accurate liver T measurement of iron overload: A simulations investigation and in vivo study

Abstract

Purpose:To investigate the accuracy of T* 2 liver iron quantification using different curve-fitting models under varying acquisition conditions, and to compare in iron-overloaded patients the reliability of rapid T* 2 measurements against approved and slower T 2 protocols. Materials and Methods:Simulations were conducted to assess the influence of various factors on the accuracy of T* 2 measurement: curve-fitting model, signal-to-noise ratio (SNR), and echo time (TE) spacing. Fifty-four iron-overloaded pediatric patients were assessed using a standard T 2 and two variations of T* 2 acquisitions. In both simulations and in vivo data, three analysis models were evaluated: monoexponential, constant offset, and truncated. Results:Simulations show the truncated model provides the best accuracy but is susceptible to underestimating high iron species under low SNR or high minimum TE. In contrast, the offset model tends to overestimate but maintains the most reliable measurements across the relevant range of iron levels. Furthermore, a much lower SNR can be tolerated if the acquisition uses a low minimum TE. In vivo results confirm theoretical findings and show that T* 2 measurements can be as reliable as those from approved and slower T 2 protocols. Conclusion:Guidelines are provided on choosing an appropriate model under specific noise conditions and acquisition schemes to ensure accurate and rapid T* 2 liver iron quantification.

Comparison between different software programs and post-processing techniques for the MRI quantification of liver iron concentration in thalassemia patients

“…bile, blood or fat) [13]. Therefore, this model is considered to be more accurate if there is any background signal from non-iron containing tissue within the ROI [13]. If there is no significant contribution to the overall signal intensity from bile, blood or fat, the truncated monoexponential decay model would be more appropriate [31].…”

“…One of the key determinants is the presence of any other signal contribution, besides iron, within the region of analysis and, in the exponential-plus-constant model, the constant offset is added to account for noise and long T2* components (i.e. bile, blood or fat) [13]. Therefore, this model is considered to be more accurate if there is any background signal from non-iron containing tissue within the ROI [13].…”

“…Relaxometry methods can be used to calculate R2 [11] or R2* values [12]-the reciprocals of T2 and T2*-by fitting a decay model to the average signal intensity at various echo times [13]. Signal intensity ratio techniques depending on T2 or T2* contrast [7][8][9] have been proposed in various studies as an immediate method for the assessment and quantification of liver iron.…”

“…Analysis of the T2* decay curves can be performed using various decay models, including the single exponential (also called monoexponential) decay and non-monoexponential models [13]. The latter category comprises the monoexponential-plus-constant offset model [12,18,19], which provides correction for noise and other artifacts in later echo images, and the truncated model, which allows the operator to manually discard from data fitting the echo times below the noise level [19,20].…”

Visualization and quantification of simian immunodeficiency virus-infected cells using non-invasive molecular imaging

“…Ferritin is primarily responsible for the storage of intracellular iron in a non-toxic form, and is a natural source of intrinsic MRI contrast in various tissues and organs (Genove et al, 2005). The level of ferritin and its iron content can be quantified by MRI, which has been widely used clinically to measure tissue iron in the human brain (Kirsch et al, 2009;Pfefferbaum et al, 2009;Yao et al, 2009), liver (Beaumont et al, 2009) and heart (Guo et al, 2009). Compared with PET imaging, MRI can provide a much higher spatial resolution, but at the sacrifice of lower sensitivity.…”